Safety Classification
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
87F1 Score
14
May 29, 2026
92F1 Score
14
Feb 26, 2026
0.96F1 Score
14
Feb 26, 2026
94Recall
14
Feb 26, 2026
100Recall
14
Feb 26, 2026
76.7F1 Score
12
Jun 25, 2026
92.96Accuracy
12
Jun 2, 2026
89.97Accuracy
12
Jun 2, 2026
99.9F1 Score
12
Jun 26, 2026
100F1 Score
12
Apr 23, 2026
100Policy L1 Success Rate
11
Mar 4, 2026
72.88F1 Score
11
Feb 26, 2026
75.8F1 Score
10
Jun 1, 2026
94.6F1 Score
10
Jun 1, 2026
94.01Accuracy
9
Jun 23, 2026
99.46Accuracy
9
Jun 23, 2026
89.48Accuracy
9
Jun 23, 2026
86.5F1 Score
9
Feb 26, 2026
93.2F1 Score
9
Feb 26, 2026
83.1F1 Score
8
Jun 1, 2026
66.98AUROC
8
Feb 26, 2026
0.936Recall
8
Feb 26, 2026
96.26Accuracy (Harmless)
7
Apr 10, 2026
100Accuracy
7
Mar 5, 2026
100Accuracy
7
Mar 5, 2026